RESEARCH Open Access

Chemokine CXCL13 acts via CXCR5-ERKsignaling in hippocampus to induceperioperative neurocognitive disorders insurgically treated miceYanan Shen1, Yuan Zhang1, Lihai Chen1, Jiayue Du1, Hongguang Bao1, Yan Xing2, Mengmeng Cai1 andYanna Si1*

Abstract

Background: Perioperative neurocognitive disorders (PNDs) occur frequently after surgery and worsen patientoutcome. How C-X-C motif chemokine (CXCL) 13 and its sole receptor CXCR5 contribute to PNDs remains poorlyunderstood.

Methods: A PND model was created in adult male C57BL/6J and CXCR5−/− mice by exploratory laparotomy. Micewere pretreated via intracerebroventricular injection with recombinant CXCL13, short hairpin RNA against CXCL13or a scrambled control RNA, or ERK inhibitor PD98059. Then surgery was performed to induce PNDs, and animalswere assessed in the Barnes maze trial followed by a fear-conditioning test. Expression of CXCL13, CXCR5, and ERKin hippocampus was examined using Western blot, quantitative PCR, and immunohistochemistry. Levels ofinterleukin-1 beta (IL-1β) and tumor necrosis factor alpha (TNF-α) in hippocampus were assessed by Western blot.

Results: Surgery impaired learning and memory, and it increased expression of CXCL13 and CXCR5 in thehippocampus. CXCL13 knockdown partially reversed the effects of surgery on CXCR5 and cognitive dysfunction.CXCR5 knockout led to similar cognitive outcomes as CXCL13 knockdown, and it repressed surgery-inducedactivation of ERK and production of IL-1β and TNF-α in hippocampus. Recombinant CXCL13 induced cognitivedeficits and increased the expression of phospho-ERK as well as IL-1β and TNF-α in hippocampus of wild-type mice,but not CXCR5−/− mice. PD98059 partially blocked CXCL13-induced cognitive dysfunction as well as production ofIL-1β and TNF-α.(Continued on next page)

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to thedata made available in this article, unless otherwise stated in a credit line to the data.

* Correspondence: siyanna@163.com1Department of Anesthesiology, Nanjing First Hospital, Nanjing MedicalUniversity, Nanjing 210006, People’s Republic of ChinaFull list of author information is available at the end of the article

Shen et al. Journal of Neuroinflammation (2020) 17:335 https://doi.org/10.1186/s12974-020-02013-x

(Continued from previous page)

Conclusions: CXCL13-induced activation of CXCR5 may contribute to PNDs by triggering ERK-mediated productionof pro-inflammatory cytokines in hippocampus.

Keywords: Perioperative neurocognitive disorders, Hippocampus, CXCL13, CXCR5, p-ERK

BackgroundPerioperative neurocognitive disorders (PNDs) refer topre- and post-operative reduction in hippocampus-dependent cognitive functions involving disturbance ofmemory, consciousness and attention, as well as changesin personality [1, 2]. PNDs are a well-recognized clinicalcomplication and patients of any age undergoing surgicalprocedure are susceptible. Among patients undergoingmajor noncardiac surgery, PNDs occur before dischargein 37% of patients aged 18–39 years, 30% of patientsaged 40–59, and 41% of patients 60 years or older [3].PNDs not only increase the burden on the healthcaresystem but also lead to postoperative long-term disabilityand even mortality. The use of volatile anesthetics dur-ing surgery has long been suspected of increasing risk ofPNDs [4, 5]. As the number of surgeries continues to in-crease, PNDs are expected to become an epidemic [1].Increasing evidence suggests that PNDs involve neuroin-flammation in the hippocampus, at least in their earlystages [2, 4]. Under certain conditions, secreted chemo-kines contribute to neuroinflammation in the centralnervous system (CNS) [6–8]. Whether and how chemo-kines may contribute to PNDs is unclear, particularly inthe hippocampus.C-X-C motif chemokine (CXCL) 13 recruits B cells to

the CNS in different neuroinflammatory diseases [6]. C-X-C motif chemokine 13 (CXCL13) helps mediate theinflammatory response to infection and injury, and it ex-erts local, concentration-dependent effects on hippo-campal memory function [9]. Elevated levels of CXCL13have been reported in the cerebrospinal fluid of patientswith multiple sclerosis or neuromyelitis optica, and ithas even been proposed as a biomarker for these neuro-degenerative conditions [7]. The only known CXCL13receptor is C-X-C motif chemokine receptor 5 (CXCR5),which is highly expressed on B lymphocytes and certainT cells in cerebrospinal fluid, and which helps guidelymphocyte tracking within secondary lymphoid tissues[10–13]. Our previous work showed that CXCR5 con-tributes to learning and memory impairment in an ani-mal model of sepsis [14]. CXCR5 is expressed inneuronal precursor cells of the brain, which migrateacross brain endothelial cells upon exposure to CXCL13[11]. CXCL13/CXCR5 signaling in the spinal cord medi-ates neuroinflammation induced by neuropathic pain[13]. Expression of CXCL13 and CXCR5 is altered in in-tractable temporal lobe epilepsy patients and epileptic

rats that show neurodegeneration and cognitive impair-ment [9]. These considerations led us to ask whetherCXCL13 and CXCR5 are involved in PNDs.We began to investigate this question by focusing on

extracellular signal-regulated kinase (ERK). The CXCL13/CXCR5 axis acts via ERK to trigger production of pro-inflammatory cytokines induced by orofacial neuropathicpain [13], and CXCL13 appears to activate ERK by trigger-ing its phosphorylation to p-ERK [15, 16]. ERK is highlyexpressed in the neurons of the developing and adultbrain, functioning as an important mediator of synapticplasticity and cell survival [17]. Changes in hippocampallevels of p-ERK have been linked to cognitive impairmentin a mouse model of Alzheimer’s disease [18], and p-ERKappears to induce production of pro-inflammatory cyto-kines, such as tumor necrosis factor alpha (TNF-α) andinterleukin-1 beta (IL-1β), in animal models of cognitivedeficits [19].The purpose of the present study was to examine

whether CXCL13 may contribute to PNDs in an animalmodel, and more concretely whether it does so via ERK.

MethodsAnimalsThe animal protocol was approved by the InstitutionalAnimal Ethics Committee of Nanjing Medical Univer-sity. All experiments described here were conducted instrict accordance with institutional guidelines.Male C57BL/6J wild-type (WT) mice (3 months old,

33–40 g) were obtained from the Experimental AnimalCenter of Nanjing First Hospital. CXCR5−/− mice[B6.129S2 (Cg)-CXCR5 tm1Lipp/J, stock number 006659]bred on a B6 background were kindly provided by Profes-sor Yongjing Gao (Pain Research Laboratory, NantongUniversity, China) [13]. WT and CXCR5−/− mice wereage-, gender-, and weight-matched. All mice were housedin a pathogen-free facility in the Experimental AnimalCenter at Nanjing First Hospital and maintained under a12-h light-dark cycle with free access to food and water.

Experimental designThe overall experimental design is summarized in Fig. 1.The mouse PND model was established using exploratorylaparotomy. All pharmacological treatments were givenvia intracerebroventricular (i.c.v.) injection into the leftventricle. Several experimental procedures with differenttimelines were performed in order to comprehensively

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 2 of 16

assess the roles of CXCL13, CXCR5, and ERK in PNDs.The following is a brief description of the experimentalgroups, and their manipulations are described in detail insubsequent subsections. (1) A cohort of WT mice was di-vided into two groups, one of which underwent laparot-omy (n = 8) and the other of which did not (n = 8; termedcontrols). No additional pre-treatments were adminis-tered. (2) A cohort of WT mice was divided to randomlyreceive short hairpin RNA (shRNA) targeting CXCL13 orscrambled control shRNA at 3 days before surgery (n = 8).Additionally, ERK inhibitor PD98059 or vehicle wasinjected 1 h before surgery (n = 8). (3) A cohort of WTand CXCR5−/− mice was randomly selected to undergosurgery without prior treatment (n = 8 of each strain). (4)

A cohort of WT and CXCR5−/− mice that did not undergosurgery was injected i.c.v. with recombinant murineCXCL13 (to mimic surgical insult) or with vehicle (n = 8).(5) A cohort of WT mice was injected with ERK inhibitorPD98059 at 1 h before treatment with recombinantCXCL13 (n = 8).

Exploratory laparotomy modelExploratory laparotomy was performed on mice in a sterileenvironment as described [20], with some modifications. Toensure a similar degree of stimulation, every mouse was ex-plored using the same set of procedures. Briefly, mice wereanesthetized by 1.5% isoflurane inhalation and kept underspontaneous respiration with an inspiration of 50% O2.

Fig. 1 Schematic illustration of the experimental design. CXCL13 C-X-C motif chemokine 13, CXCR5 C-X-C motif chemokine receptor 5, HE hematoxylin-eosinstaining, i.c.v. intracerebroventricular, IL-1β interleukin-1 beta, TNF-α tumor necrosis factor alpha,WT wide-type

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 3 of 16

Temperature was monitored rectally (Techman Software,Chengdu, China) and maintained at 37 ± 0.5 °C with theaid of a heating blanket. To mimic clinical exploratorylaparotomy, a longitudinal midline incision was made inthe abdomen from the xiphoid process to the superiormargin of the pubic symphysis. A cotton tip wetted with0.9% sterile saline (Baxter Healthcare, Shanghai, China)explored abdominal organs in the following order: liver,spleen, left and right kidneys, bladder and bowel. The inci-sion site was infiltrated with bupivacaine (0.25%, 3 mg/kg)for postoperative analgesia. Then, the peritoneum andskin were sutured separately. The total duration of explor-ation and anesthesia lasted around 2 h. After animals re-covered, they were returned to their cages with food andwater ad libitum.

Pharmacological treatmentsAll drugs were prepared under sterile conditions. Thedose for i.c.v. injection was 5 μg for PD98059 (Sigma, St.Louis, MO, USA) [17] as well as recombinant murineCXCL13 (catalog no. 250-24, PeproTech, Rocky Hills,NJ, USA) [13]. The final DMSO concentration was 2.5%.Vehicle solution was 2.5% DMSO. Solutions wereadministered by i.c.v. injection. An shRNA targetingCXCL13 was expressed from the sequence 5′-TCGTGC CAA ATG GTT ACA A-3′ in a plasmid (Gene-Pharma, Shanghai, China) [13], while negative controlscrambled shRNA was expressed from the sequence 5′-TTC TCC GAA CGT GTC ACG T-3′. The shRNAs (5μg) were dissolved in 5 μL RNase-free water, mixed with5 μL transfection reagent (Entranster™-in vivo, EngreenBiosystem, Beijing, China), and injected i.c.v. at 3 daysbefore exploratory laparotomy as described previously[14]. CXCL13 knockdown was verified using Westernblot analysis.

Intracerebroventricular injectionsMice were anesthetized with 1.5% isoflurane inhalation,and placed in a stereotaxic apparatus (Stoelting, WoodDale, IL, USA). I.c.v. injection was performed using a 32-gauge microsyringe (Hamilton, Reno, NV, USA) over aperiod of 10 min, and the microsyringe was left in placefor 5 min after completing the infusion. The stereotaxiccoordinates were determined as 1.0 mm to the left frommidline, 0.5 mm posterior to the bregma, and 2.5 mmbelow the dural surface [21]. The animals were taken tothe individual cages after the i.c.v. injection.

Barnes maze testThe Barnes maze trial was performed to assess thespatial learning and memory of mice as described [22],with modifications. Animals were placed in the middleof a round platform with 20 equal-sized peripheral holesspaced at regular intervals. One of the holes, called the

“target hole,” was connected to a dark box. After themouse was placed in the middle of the platform, aversivenoise (85 dB) and bright light from a 200-W bulb wereused to stimulate mice to seek and enter the target hole.Mice were trained in a spatial acquisition phase (3 minper trial, 2 trials each day at a 2-h interval for 4 days)beginning on day 7 after surgery. The spatial memorytest was conducted 1 day after training ended (day 11post-surgery) to measure short-term retention and 8days after training (day 18 after surgery) to measurelong-term retention. Mouse activity and the latency tofind the target hole during each trial were recordedusing the ANY-Maze video tracking system (SD Instru-ments, San Diego, CA, USA).

Fear conditioning testA fear conditioning test was conducted as described [23]on days 19 and 20 after surgery. All mice were placed ina chamber with a floor of stainless steel bars (LE111,Panlab, Barcelona, Spain) that had been wiped downwith 70% ethanol. This chamber was designed to allowsound stimulation for the training of conditionedreflexes, as well as electrical stimulation on the feetusing a constant-current generator. Mice were allowedto explore and habituate in the training environment for3 min. Then they were stimulated with a monofrequencysound for 30 s (1 kHz, 80 dB), and during the last 2 s ofthis sound, a foot shock (1 mA) was delivered. After a30-s pause, this pairing of sound and shock were deliv-ered a second time. After 24 h, as a test of contextualmemory, mice were allowed to enter the original cham-ber without any stimulation, and freezing behavior,defined as the absence of any movement other than thatrequired for respiration for more than 3 s, was moni-tored using a SMART3.0 video tracking system (Panlab).Then, 2 h later, a cued fear conditioning test was con-ducted in which the animals were placed in the chamberand subjected to the monofrequency sound for 3 minwithout foot shock, and freezing behavior was monitoredand recorded.

Western blotHippocampus tissues were homogenized and centri-fuged at 13,000 g at 4 °C for 10 min. The supernatantwas separated on SDS-PAGE gels, and then trans-ferred to a nitrocellulose membrane (Hybond-ECL,Amersham Biosciences, Little Chalfont, UK). Mem-branes were blocked, then incubated overnight at 4°C with antibodies against CXCL13 (1:100; NovusBiologicals, Littleton, CO, USA) or CXCR5 (1:1000;Santa Cruz Biotechnology, Santa Cruz, CA, USA); orantibodies against ERK (1:500), p-ERK (1:500), IL-1β(1:1000), or TNF-α (1:500) (Cell Signaling, Beverly,MA, USA). Next, blots were incubated with secondary

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 4 of 16

antibodies (Sigma-Aldrich) conjugated to horseradishperoxidase. Protein bands were detected using en-hanced chemiluminescence (Amersham Biosciences)and quantitated using Quantity One software (Bio-Rad Laboratories, Hercules, CA, USA). Band inten-sities were normalized to that of β-actin.

Reverse transcription-quantitative polymerase chainreactionTotal RNA was extracted from hippocampal tissue usingTrizol (Invitrogen, Carlsbad, CA, USA). Total RNA (1 μg)was reverse-transcribed using a reverse transcription kit (no.DRR047A; Takara Bio, Otsu, Japan) according to the manu-facturer’s protocol. Reverse transcription-quantitative poly-merase chain reaction (RT-qPCR) was performed usingSYBR Green I dye detection (Takara Bio) in a Real-Time De-tection system (Bio-Rad Laboratories). The following primerswere used [13]: CXCL13 forward, 5′-GGC CAC GGT ATTCTG GAA GC-3′; CXCL13 reverse, 5′-ACC GAC AACAGT TGA AAT CAC TC-3′; CXCR5 forward, 5′-TGGCCT TCTA CAG TAA CAG CA-3′; CXCR5 reverse, 5′-GCA TGA ATA CCG CCT TAA AGG AC-3′; GAPDHforward, 5′-GCT TGA AGG TGT TGC CCT CAG-3′; andGAPDH reverse, 5′-AGA AGC CAG CGT TCA CCAGAC-3′. PCR conditions included an initial step at 95 °C for3 min, followed by 30 cycles at 95 °C for 1 min, 56 °C for 40s, and 72 °C for 1 min. Melting curves were generated to ver-ify amplification specificity. Quantification was performedusing the 2-ΔΔCT method, and cycle threshold values werenormalized to those of GAPDH.

Hematoxylin and eosin stainingMice were perfused with 20-mL normal saline,followed by 20-mL 4% formaldehyde. Brain tissueswere harvested and post-fixed with 4% paraformalde-hyde at room temperature overnight and then em-bedded in paraffin. A series of 5-μm-thick coronalsections were deparaffinized, rehydrated, and stainedby hematoxylin-eosin (HE). The stained sectionswere observed under an optical microscope (Olym-pus BX53, Tokyo, Japan) in order to assess histo-pathology in the hippocampal region.

Immunohistochemical stainingBrain tissue was immunostained as described [24]. It was iso-lated, fixed in 4% paraformaldehyde, placed into paraffinblocks, and cut coronally into 6-μm sections. Sections weredeparaffinized, rehydrated, heated at 95 °C for 15 min forantigen retrieval, then incubated in 3% H2O2 for 25 min atroom temperature to block endogenous peroxidases. Afterblocking by goat serum for 30 min, the sections were incu-bated overnight at 4 °C with primary antibodies againstCXCL13 (1:100; ab272874, Abcam, Cambridge, UK) orCXCR5 (1:100; ab133706, Abcam) or p-ERK (1:250; 4695,

CTS, Danvers, MA, USA). Sections were then incubatedwith secondary antibodies for 60 min at room temperature.A 3,3′-diaminobenzidine kit (Bioss, Beijing, China) was usedto visualize antibody binding, and the sections were counter-stained with hematoxylin. Photomicrographs were capturedby a digital camera connected to a microscope (3D HISTECH, Budapest, Hungary). The immunoreactive cells inCA1, CA3, and dentate gyrus of hippocampus were countedusing Image Pro-Plus 6.0 (Media Cybernetics, Bethesda, MD,USA) in three randomly selected fields on each section.

Statistical analysisStatistical analysis was carried out using GraphPad PrismVersion 5.01 (Graph Pad Software, San Diego, CA, USA).Continuous variables were assessed using Shapiro-Wilktest for distribution. The analysis showed normal distribu-tion for all variables. Thus, all continuous variables areshown as mean ± standard deviation (SD). Inter-group dif-ferences in Barnes maze training results were assessed forsignificance using two-way repeated-measures ANOVA,followed by the Bonferroni post hoc multiple comparisontest. Inter-group differences on other variables wereassessed using Student’s t test (for comparisons betweentwo groups) or one-way ANOVA followed by Tukey’smultiple test (for comparisons among at least threegroups). P < 0.05 was considered statistically significant.

ResultsSurgery in mice impairs cognitive function andhippocampal morphology as a model of PNDsTimes in the Barnes maze decreased over the four trainingsessions in both control and surgery groups, and the micethat underwent surgery took longer to identify the target holethan controls (Fig. 2a). Surgery significantly prolonged la-tency time to identify the target hole at 1 or 8 days aftertraining (Fig. 2b). Surgery shortened freezing times in thecontextual fear conditioning test (Fig. 2c) without affectingfreezing times in the cued fear conditioning test (Fig. 2d). HEstaining also showed that surgery resulted in irregular neur-onal arrangement, loss of morphology, and increases in neur-onal swelling, nuclear pyknosis, and cellular vacuolization inthe hippocampus (Fig. 2e).

Surgery upregulates CXCL13 in mouse hippocampusWe checked CXCL13 mRNA levels in the hippocam-pus on days 1, , and 7 after surgery. Tissue from sur-gically treated mice had higher CXCL13 mRNA levelsthan control tissue at all timepoints (Fig. 3a).CXCL13 protein expression was higher in surgicallytreated animals than in controls on day 7 aftersurgery (Fig. 3b). Immunohistochemical staining ofhippocampus showed more CXCL13-expressing neu-rons in the CA1, CA3, and dentate gyrus at 7 daysafter surgery than in controls (Fig. 3c).

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 5 of 16

CXCL13 knockdown alleviates PND-like cognitive deficitsin miceMice that underwent surgery alone showed higherCXCL13 expression than controls (Fig. 4 a, b). To exam-ine the role of CXCL13 in the development and

maintenance of cognitive dysfunction, we injectedCXCL13 shRNA into the lateral ventricle 3 days beforesurgery. CXCL13 knockdown significantly reducedCXCL13 expression in surgically treated mice, whilescrambled RNA had no effect on CXCL13 expression (Fig.

Fig. 2 Surgery in mice impairs cognitive function and hippocampal morphology as a model of PNDs. Adult male C57BL/6J mice were subjectedto exploratory laparotomy (termed “Surgery”). Control mice did not undergo surgery. All mice were subjected to 4 days of training in the Barnesmaze beginning 7 days after the surgery; memory phase testing was performed on days 11 and 18 post-surgery. a Time to find the target holeduring the training sessions and b latency in identifying the target hole after training were measured. Fear conditioning tests were performed ondays 19 and 20 after surgery. Freezing times were quantified c in context and d in response to a cue (n = 8 per group). e Micrographs ofhippocampal sections stained with hematoxylin and eosin (n = 4 per group). Magnification, × 20 for regions CA1, CA3, and DG in hippocampus.*P < 0.05 vs. control

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 6 of 16

4 a, b). Time to identify the target hole in the Barnes mazetrial progressively decreased over the 4 days of trainingsessions in all groups (Fig. 4c). All surgically treatedgroups took significantly longer to find the targethole than the control group during training and test-ing, regardless of CXCL13 levels (Fig. 4 c, d ). How-ever, mice transfected with CXCL13 shRNA prior tosurgery identified the target hole faster than animalstreated with surgery alone or with surgery andscrambled shRNA.All mice that underwent surgery exhibited shorter

freezing times than controls in the contextual fear con-ditioning test (Fig. 4e), and CXCL13 knockdown par-tially reversed this surgery-induced impairment. Incontrast, the groups did not differ significantly in freez-ing times in the hippocampal-independent cued fearconditioning test (Fig. 4f).

Surgery upregulates CXCR5 in mouse hippocampusCXCR5 mRNA levels were higher in hippocampal tis-sue at 1, 3, and 7 days after surgery (Fig. 5a). Similarly,the level of corresponding protein was higher in surgi-cally treated mice than controls at 7 days after surgery,as shown by Western blot and immunohistochemistry(Fig. 5 b, c ).

Deletion of CXCR5 alleviates PND-like cognitive deficits in miceWhile all groups became increasingly faster with eachtraining session in the Barnes maze trial, surgery im-pacted time more than CXCR5 levels (Fig. 6a). Surgicallytreated CXCR5−/− mice identified the target hole fasterthan surgically-treated WT mice (Fig. 6b).Although all mice that underwent surgery exhibited

shorter freezing times than control mice in the contextualfear conditioning test, no differences were seen among the

Fig. 3 Surgery up-regulates CXCL13 in mouse hippocampus. a Relative level of CXCL13 mRNA was measured by RT-qPCR in wild type mice at 1,3, and 7 days after surgery. b Western blotting analysis of CXCL13 at 7 days after surgery. β-actin was used as an internal control (n = 8 pergroup). c Representative images of CXCL13 immunohistochemistry in the hippocampal CA1, CA3, and dentate gyrus regions. Positively stainedcells are brown (red arrow). Magnification, × 5 for total hippocampus; × 20 for regions CA1, CA3, and DG in hippocampus. All experiments wereperformed with n = 4 per group. *P < 0.05 vs. control. CXCL13 C-X-C motif chemokine 13, DG dentate gyrus, RT-qPCR reverse transcription-quantitative polymerase chain reaction

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 7 of 16

Fig. 4 CXCL13 knockdown alleviates PND-like cognitive deficits in mice. Mice were pretreated with CXCL13 shRNA or scrambled shRNA, subjected tosurgery and then behavioral tests were performed 7 days after surgery. a Western blotting and densitometry of hippocampal CXCL13, as well as b RT-qPCR analysis of the corresponding mRNA. c Time to identify the target hole in training sessions and d latency time during memory phases werequantified in the Barnes maze. In the fear conditioning tests, freezing times were quantified e in context and f in response to a cue. All quantitativeresults (mean ± SD) shown were obtained with n = 8 animals per group. *P < 0.05 vs. control; #P < 0.05 vs. surgery. CXCL13 C-X-C motif chemokine 13,RT-qPCR reverse transcription-quantitative polymerase chain reaction, shRNA short hairpin RNA, sh-CRTL scrambled shRNA, sh-CXCL13 CXCL13 shRNA

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 8 of 16

groups during the cued fear conditioning test (Fig. 6 c, d ).Surgically treated WT mice had higher CXCR5 mRNAand protein expression than control WT mice (Fig. 6 e, f ).

CXCR5 deficiency inhibits surgery-induced p-ERKactivation and release of inflammatory cytokines inmouse hippocampusWT and CXCR5−/− mice with surgery had significantlyhigher p-ERK expression in the hippocampus on day 7after surgery than counterparts that did not undergosurgery (Fig. 7a). However, total ERK levels were un-affected. CXCR5 knockout reduced p-ERK levels in sur-gically treated mice but not control mice. Moreover,surgery induced an increase in the number of p-ERK-containing neurons in the hippocampus of WT mice onday 7, and less so in CXCR5−/− mice (Fig. 7b). There wasno difference in the number of p-ERK-containing neu-rons in control animals of either genotype.

TNF-α and IL-1β are important pro-inflammatory cyto-kines in regulating cognitive dysfunction in the CNS [25].We checked TNF-α and IL-1β expression in the hippo-campus on day 7 after surgery. WT mice with surgery hadhigher TNF-α and IL-1β expression than controls (Fig.7c). Surgically treated CXCR5−/− mice had lower TNF-αand IL-1β expression than surgically treated WT mice.

ERK-dependent production of pro-inflammatory cytokines inhippocampus mediates PND-like cognitive deficits in miceMice pretreated with PD98059 displayed shorter timeto the target hole on all training days (Fig. 8a), andalso had shorter latency than surgically-treated WTmice (Fig. 8b). PD98059 pretreatment also rescuedPND-like effects in the contextual fear conditioningtest (Fig. 8c), but made no difference in cued fearfreezing (Fig. 8d). PD98059 pretreatment downregu-lated TNF-α and IL-1β expression in surgicallytreated mice (Fig. 8e).

Fig. 5 Surgery up-regulates CXCR5 in mouse hippocampus. a Relative level of CXCR5 mRNA was measured by RT-qPCR on days 1, 3, and 7 after surgery.bWestern blotting analysis and densitometry of CXCR5 protein levels. β-actin was used as an internal control (n = 8 per group). c Representative images ofCXCR5 immunohistochemistry in the hippocampal CA1, CA3, and dentate gyrus. Positively stained cells are brown (red arrow). Magnification, × 5 for totalhippocampus; × 20 for hippocampal CA1, CA3, and DG (n = 4 per group). *P < 0.05 vs. control. CXCR5 C-X-C motif chemokine receptor 5, DG dentate gyrus, RT-qPCR reverse transcription-quantitative polymerase chain reaction

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 9 of 16

Fig. 6 Deletion of CXCR5 alleviates PND-like cognitive deficits in mice. WT or mice lacking the CXCR5 gene (CXCR5−/−) were subjected to surgery(denoted as “surgery”). Mice that did not undergo surgery were denoted “control.” All mice were tested in the Barnes maze and fear conditioningtests at 7 days after surgery. a Time to identify the target hole during training sessions and b latency time during memory phases of the Barnesmaze. In the fear conditioning test, freezing times were quantified c in context and d in response to a cue. e Western blot and densitometry ofhippocampal CXCR5, as well as f RT-qPCR analysis of the corresponding mRNA. All quantitative results (mean ± SD) shown were obtained with n= 8 animals per group. *P < 0.05 vs. WT + control; #P < 0.05 vs. WT + surgery. CXCR5 C-X-C motif chemokine receptor 5, RT-qPCR reversetranscription-quantitative polymerase chain reaction, WT wild-type

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 10 of 16

Fig. 7 CXCR5 deficiency inhibits surgery-induced p-ERK activation and release of inflammatory cytokines in mouse hippocampus. a Westernblotting analysis of total ERK and p-ERK (n = 8 per group). b Representative images of p-ERK immunohistochemistry in the hippocampus. Positivecells are stained brown. Magnification, × 5 for total hippocampus; × 20 for hippocampal CA1, CA3, and DG (n = 4 per group). c Western blot anddensitometry of IL-1β and TNF-α in the hippocampus. β-actin was used as an internal control (n = 8 per group). *P < 0.05 vs. control; #P < 0.05 vs.surgery. CXCR5 C-X-C motif chemokine receptor 5, DG dentate gyrus, WT wild-type

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 11 of 16

Fig. 8 ERK-dependent production of pro-inflammatory cytokines in hippocampus mediates PND-like cognitive deficits in mice. Mice subjected tosurgery were pretreated with vehicle or PD98059, and behavioral tests were conducted 7 days after the surgery. a Time to find the target hole duringtraining sessions and b latency time during memory phases were assessed in the Barnes maze. In the fear conditioning test, freezing times werequantified c in context and d in response to a cue. e Western blot and densitometry of hippocampal IL-1β and TNF-α protein levels. β-actin was usedas an internal control. All quantitative results (mean ± SD) shown were obtained with n = 8 animals per group. *P < 0.05 vs. control; #P < 0.05 vs.surgery. TNF-α tumor necrosis factor alpha, IL-1β interleukin-1 beta

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 12 of 16

CXCL13 induces CXCR5/ERK-dependent release of pro-inflammatory cytokines in mouse hippocampusCompared with control mice, all CXCL13 treatmentgroups took more time to identify the target hole acrossall days (Fig. 9 a, b ), and these impairments were par-tially reversed by deleting CXCR5. Additionally, CXCL13treatment shortened freezing time in contextual fearconditioning, and it did so to a smaller extent inCXCR5−/− mice (Fig. 9c). No differences were found inthe cued fear conditioning test (Fig. 9d). CXCL13 in-creased the levels of p-ERK, IL-1β, and TNF-α, and theeffect was mitigated by CXCR5 deficiency (Fig. 9e).Pretreatment PD98059 partially attenuated CXCL13-

induced cognitive dysfunction (Fig. 9 f–i ). Additionally,CXCL13 mice pretreated with PD98059 had lower

expression of TNF-α and IL-1β than mice treated withCXCL13 alone (Fig. 9j).

DiscussionThe present study explored whether chemokine CXCL13and its receptor CXCR5 are involved in surgery-inducedcognitive dysfunction in an animal model of PND. Ourdata clearly demonstrate that surgical intervention in-duced CXCL13 release and an increase in CXCR5 ex-pression followed by cognitive impairment as observedin behavioral tests, and that inhibition of CXCL13/CXCR5 signaling attenuated surgery-induced cognitiveimpairment. We also investigated the role of ERK insurgery-induced neuroinflammation. Surgery inducedERK activation and ERK-dependent production of the

Fig. 9 CXCL13 induces CXCR5/ERK-dependent release of pro-inflammatory cytokines in mouse hippocampus. WT or CXCR5−/− mice receivedCXCL13 via intracerebroventricular injection. Seven days later, Barnes maze and fear conditioning tests were conducted. a Time to identify thetarget hole during training sessions and b latency time during memory phase were measured in the Barnes maze. Freezing times were quantifiedc in context and d in response to a cue. e Western blotting and densitometry of hippocampal p-ERK, total ERK, IL-1β, and TNF-α. Mice wereinjected with CXCL13 with or without PD98059 pretreatment, and behavioral tests were performed 7 days later. f Time to identify the target holein training sessions and h latency time in memory phases of the Barnes maze. Freezing times were quantified i in context and j in response to acue in fear conditioning tests. k Western blot and densitometry of IL-1β and TNF-α. All quantitative results (mean ± SD) shown were obtainedwith n = 8 animals per group. *P < 0.05 vs. vehicle + WT; #P < 0.05 vs. CXCL13 + WT; $P < 0.05 vs. vehicle; &P < 0.05 vs. CXCL13. CXCL13 C-X-Cmotif chemokine 13, CXCR5 C-X-C motif chemokine receptor 5, WT wild-type, TNF-α tumor necrosis factor alpha, IL-1β interleukin-1 beta

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 13 of 16

inflammatory factors IL-1β and TNF-α in the hippocam-pus of mice, but this effect was repressed by CXCR5knockout. Finally, i.c.v. injection of recombinantCXCL13 induced CXCR5/ERK-dependent upregulationof TNF-α and IL-1β as well as cognitive dysfunction.Altogether, our results suggest that CXCL13 inducesPND, at least partially, by triggering hippocampalCXCR5-ERK signaling.CXCL13, a member of the chemokine family, is the

major determinant for B cell recruitment to the intra-thecal compartment during a range of human neuro-logical disorders [6, 26]. Studies in animals and humanssuggest that CXCL13 contributes to certain CNS disor-ders, and that blockade of this ligand might have thera-peutic effects [25]. CXCL13 has been proposed as abiomarker of neuroinflammatory disease due to its highlevel in the CNS under a variety of conditions, includingmultiple sclerosis and neuromyelitis optica [7] as well asCNS lymphoma [27, 28]. Whether CXCL13 is involvedin the pathogenesis of PNDs remains unclear. In thepresent study, CXCL13 mRNA in the hippocampus wasupregulated 1 day after surgery and remained elevatedfor more than 7 days in surgically treated mice with cog-nitive impairment; Western blotting and immunostain-ing further showed upregulated CXCL13 protein in thehippocampus. In agreement with our results, several re-ports showed that CXCL13 was increased in dorsal rootganglion in a neuropathic pain model and hippocampusin CNS infection [29–32], and its high expression wasdetected mainly in microglia, macrophages, and endo-thelial cells [13, 33]. To examine the role of CXCL13 inthe development and maintenance of cognitive dysfunc-tion, we knocked it down specifically using shRNA.Knockdown effectively attenuated surgery-induced cog-nitive dysfunction. These data suggest the involvementof hippocampal CXCL13 in surgery-induced PNDs.CXCR5 is responsible for the organization of B cell

follicles and for directing T-helper cells to the lymphoidfollicles [34–37]. In the CNS, CXCR5 is expressed inneuronal precursor cells, which migrate across brainendothelial cells upon exposure to CXCL13 [11].CXCL13 and CXCR5 are involved in neurodegenerationand cognitive decline [9], which are related to a deficitin hippocampal neurogenesis [38, 39]. In previous work,we showed that CXCR5 may inhibit proliferation, differ-entiation, and survival of hippocampal neuronal stemcells in an animal model of sepsis, leading to learningand memory impairments [14]. Hippocampal CXCR5upregulation causes sevoflurane anesthesia-induced cog-nitive dysfunction via PI3K/Akt signaling [40]. In thepresent study, we found that surgery elevated hippocam-pal CXCR5 mRNA and protein levels for more than 7days in mice. CXCR5 deficiency attenuated surgery-induced cognitive dysfunction. Additionally, i.c.v.

injection of recombinant CXCL13 triggered learning andmemory deficits in surgically treated animals, whichwere attenuated in CXCR5−/− animals. These data sug-gest that CXCL13/CXCR5 signaling is necessary for thedevelopment and maintenance of PNDs.CXCL13/CXCR5 signaling contributes to neurodegen-

eration in individuals with cognitive deficit disease [9], aswell as in animal models with neuropathic pain [41]. Infact, animal studies have shown that CXCL13 and CXCR5contribute to neuropathic pain via ERK signaling [13].ERK is also activated in the hippocampus of animals anes-thetized with sevoflurane or propofol, and these animalslater show impaired learning and memory [42, 43]. Intra-prefrontal cortex administration of ERK inhibitorPD98059 in 6-month-old 3xTg mice reversed memoryimpairment, suggesting that ERK pathway alterationsmight at least partially explain memory deficits observedin an Alzheimer’s disease model [44]. Here, we found thatsurgery increased p-ERK levels in the hippocampus, andthis upregulation was partially reversed by deletingCXCR5. Moreover, i.c.v. injection of CXCL13 inducedCXCR5-dependent p-ERK upregulation in the hippocam-pus. Similarly, a recent study showed that CXCL13 acti-vates ERK in the spinal cord through CXCR5 in a mousemodel of diabetes-induced tactile allodynia [41]. Thesedata suggest the important role of ERK in mediatingCXCL13/CXCR5 signaling in the hippocampus.Surgical trauma can cause neuroinflammation involving

synaptic deficits, neuronal dysfunction and death, and im-paired neurogenesis [45, 46]. ERK is known to mediatethe expression of inflammatory mediators, includinggrowth factors and pro-inflammatory cytokines (e.g., IL-1β) in an animal model of cognitive deficit [47]. IL-1β andTNF-α are important pro-inflammatory cytokines thatmediate surgery-induced cognitive dysfunction [48]. Wealso found that surgery upregulated the pro-inflammatorycytokines IL-1β and TNF-α in the hippocampus, and thiswas partially reversed by deleting CXCR5 or knockingdown ERK. Pretreatment with CXCL13 induced CXCR5/ERK-dependent release of pro-inflammatory cytokines inthe hippocampus. Both CXCL13 and CXCR5 were upreg-ulated in the brain in both patients with intractable epi-lepsy and a rat model of epilepsy with neuroinflammation[9]. Increased release of proinflammatory cytokines, in-cluding IL-1β and TNF-α, has been reported in the hippo-campus of rodents after prolonged seizures [49], andcontributes to cognitive impairment [50]. We did not ob-serve spontaneous seizure in our experiments. Such a dis-crepancy warrants further investigation. Nevertheless, ourfindings data suggest that ERK-mediated release of pro-inflammatory cytokines in the hippocampus contributesto PNDs.Previous studies showed that CXCR5−/− mice have more

immature neurons in the dentate gyrus, an increase in

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 14 of 16

baseline locomotor activity, and decreased anxiety-like be-havior [39]. CXCR5−/− mice may also develop retinal degen-eration with pathophysiological changes, such as activationof microglia and accumulation of inflammatory cells, loss ofZO-1 indicative of impaired blood-retinal barrier functionafter ischemia-reperfusion injury [51]. These changes couldconceivably produce major impact on learning and memory,and thus bias the results in the current study.

ConclusionsThese results from the current study indicated thatCXCL13 and CXCR5 contribute to PNDs, possibly byactivating ERK to trigger release of pro-inflammatory cy-tokines in the hippocampus. Thus, targeting theCXCL13/CXCR5/ERK pathway in the hippocampus mayprovide a novel therapeutic approach to treat thesedisorders.

AbbreviationsCNS: Central nervous system; CXCL13: C-X-C motif chemokine 13; CXCR5: C-X-C motif chemokine receptor 5; ERK: Extracellular signal-regulated kinase;i.c.v.: Intracerebroventricular; IL-1β: Interleukin-1 beta; PND: Perioperativeneurocognitive disorder; RT-qPCR: Reverse transcription-quantitative polymer-ase chain reaction; shRNA: Short hairpin RNA; TNF-α: Tumor necrosis factoralpha; WT: Wild-type

AcknowledgementsNot applicable.

Authors’ contributionsSYN (Yanna Si) conceived and designed the study. SYN (Yanan Shen)performed the majority of the experiments. DJY and CMM assisted withtissue sample collection. ZY and XY analyzed the data. SYN (Yanna Si) andSYN (Yanan Shen) drafted and edited the manuscript. BHG and CLHprovided critical discussion and participated in the revision. All authors readand approved the final manuscript.

FundingThis work was supported by the National Natural Science Foundation ofChina (81873954 and 81971872), Nanjing Medical Science and TechnicalDevelopment Foundation (QRX17019 and YKK18105) and Six Talent PeaksProject in Jiangsu (WSW-106).

Availability of data and materialsAll data generated or analyzed during this study are available from thecorresponding author upon reasonable request.

Ethics approval and consent to participateNo human data or tissues were used in this study. All animal experimentalprotocols and handling procedures were approved by the InstitutionalAnimal Care and Use Committee of Nanjing Medical University.

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no conflicts of interest.

Author details1Department of Anesthesiology, Nanjing First Hospital, Nanjing MedicalUniversity, Nanjing 210006, People’s Republic of China. 2Jiangsu KeyLaboratory for Design and Manufacture of Micro-Nano BiomedicalInstruments, School of Mechanical Engineering, Southeast University, Nanjing211118, People’s Republic of China.

Received: 16 May 2020 Accepted: 28 October 2020

References1. Evered L, Silbert B, Knopman DS, Scott DA, DeKosky ST, Rasmussen LS, Oh

ES, Crosby G, Berger M, Eckenhoff RG, Nomenclature Consensus WorkingGrou. Recommendations for the nomenclature of cognitive changeassociated with anaesthesia and surgery-2018. Br J Anaesth. 2018;121(5):1005–12.

2. Fan D, Li J, Zheng B, Hua L, Zuo Z. Enriched environment attenuatessurgery-induced impairment of learning, memory, and neurogenesispossibly by preserving BDNF expression. Mol Neurobiol. 2016;53(1):344–54.

3. Monk TG, Weldon BC, Garvan CW, Dede DE, van der Aa MT, Heilman KM,Gravenstein JS. Predictors of cognitive dysfunction after major noncardiacsurgery. Anesthesiology. 2008;108(1):18–30.

4. Huang C, Irwin MG, Wong GTC, Chang RCC. Evidence of the impact ofsystemic inflammation on neuroinflammation from a non-bacterialendotoxin animal model. J Neuroinflammation. 2018;15(1):147.

5. Zheng B, Lai R, Li J, Zuo Z. Critical role of P2X7 receptors in theneuroinflammation and cognitive dysfunction after surgery. Brain BehavImmun. 2017;61:365–74.

6. Kowarik MC, Cepok S, Sellner J, Grummel V, Weber MS, Korn T, et al. CXCL13is the major determinant for B cell recruitment to the CSF duringneuroinflammation. J Neuroinflammation. 2012;9:93.

7. Alvarez E, Piccio L, Mikesell RJ, Klawiter EC, Parks BJ, Naismith RT, et al.CXCL13 is a biomarker of inflammation in multiple sclerosis, neuromyelitisoptica, and other neurological conditions. Mult Scler. 2013;19(9):1204–8.

8. Bagaeva LV, Rao P, Powers JM, Segal BM. CXC chemokine ligand 13 plays arole in experimental autoimmune encephalomyelitis. J Immunol. 2006;176(12):7676–85.

9. Li R, Ma L, Huang H, Ou S, Yuan J, Xu T, et al. Altered expression of CXCL13and CXCR5 in intractable temporal lobe epilepsy patients and pilocarpine-induced epileptic rats. Neurochem Res. 2017;42(2):526–40.

10. Krumbholz M, Theil D, Cepok S, Hemmer B, Kivisäkk P, Ransohoff RM, et al.Chemokines in multiple sclerosis: CXCL12 and CXCL13 up-regulation isdifferentially linked to CNS immune cell recruitment. Brain. 2006;129(Pt 1):200–11.

11. Weiss N, Deboux C, Chaverot N, Miller F, Baron-Van Evercooren A, CouraudPO, et al. IL8 and CXCL13 are potent chemokines for the recruitment ofhuman neural precursor cells across brain endothelial cells. JNeuroimmunol. 2010;223:131–4.

12. Moser B, Schaerli P, Loetscher P. CXCR5(+) T cells: follicular homing takescenter stage in T-helper-cell responses. Trends Immunol. 2002;23(5):250–4.

13. Zhang Q, Cao DL, Zhang ZJ, Jiang BC, Gao YJ. Chemokine CXCL13 mediatesorofacial neuropathic pain via CXCR5/ERK pathway in the trigeminalganglion of mice. J Neuroinflammation. 2016;13(1):183.

14. Shen Y, Jing L, Zhang Y, Bao H, Vohra A, Si Y, et al. CXCR5 knockdownattenuates hippocampal neurogenesis deficits and cognitive impairment ina mouse model of sepsis-associated encephalopathy. Neuroscience. 2020;433:212–20.

15. El-Haibi CP, Singh R, Sharma PK, Singh S, Lillard JW Jr. CXCL13 mediatesprostate cancer cell proliferation through JNK signalling and invasionthrough ERK activation. Cell Prolif. 2011;44(4):311–9.

16. Bu HL, Xia YZ, Liu PM, Guo HM, Yuan C, Fan XC, et al. The roles ofchemokine CXCL13 in the development of bone cancer pain and theregulation of morphine analgesia in rats. Neuroscience. 2019;406:62–72.

17. Yang Y, Zhang X, Cui H, Zhang C, Zhu C, Li L. Apelin-13 protects the brainagainst ischemia/reperfusion injury through activating PI3K/Akt and ERK1/2signaling pathways. Neurosci Lett. 2014;568:44–9.

18. Denner LA, Rodriguez-Rivera J, Haidacher SJ, Jahrling JB, Carmical JR,Hernandez CM, et al. Cognitive enhancement with rosiglitazone links thehippocampal PPARγ and ERK MAPK signaling pathways. J Neurosci. 2012;32(47):16725–35a.

19. Jahrling JB, Hernandez CM, Denner L, Dineley KT. PPARγ recruitment toactive ERK during memory consolidation is required for Alzheimer's disease-related cognitive enhancement. J Neurosci. 2014;34(11):4054–63.

20. Pan K, Li X, Chen Y, Zhu D, Li Y, Tao G, et al. Deferoxamine pre-treatmentprotects against postoperative cognitive dysfunction of aged rats bydepressing microglial activation via ameliorating iron accumulation inhippocampus. Neuropharmacology. 2016;111:180–94.

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 15 of 16

21. Ni J, Wang X, Chen S, Liu H, Wang Y, Xu X, et al. MicroRNA let-7c-5pprotects against cerebral ischemia injury via mechanisms involving theinhibition of microglia activation. Brain Behav Immun. 2015;49:75–85.

22. Liang P, Shan W, Zuo Z. Perioperative use of cefazolin amelioratespostoperative cognitive dysfunction but induces gut inflammation in mice.J Neuroinflammation. 2018;15(1):235.

23. Ji MH, Tang H, Luo D, Qiu LL, Jia M, Yuan HM, et al. Environmentalconditions differentially affect neurobehavioral outcomes in a mouse modelof sepsis-associated encephalopathy. Oncotarget. 2017;8(47):82376–89.

24. Hu S, Hu M, Liu J, Zhang B, Zhang Z, Zhou FH, et al. Phosphorylation of Tauand α-synuclein induced neurodegeneration in MPTP mouse model ofParkinson's disease. Neuropsychiatr Dis Treat. 2020;16:651–63.

25. Huber AK, Irani DN. Targeting CXCL13 during neuroinflammation. AdvNeuroimmune Biol. 2015;6(1):1–8.

26. Campbell DJ, Kim CH, Butcher EC. Chemokines in the systemic organizationof immunity. Immunol Rev. 2003;195:58–71.

27. Smith JR, Braziel RM, Paoletti S, Lipp M, Uguccioni M, Rosenbaum JT.Expression of B-cell-attracting chemokine 1 (CXCL13) by malignantlymphocytes and vascular endothelium in primary central nervous systemlymphoma. Blood. 2003;101(3):815–21.

28. Fischer L, Korfel A, Pfeiffer S, Kiewe P, Volk HD, Cakiroglu H, et al. CXCL13and CXCL12 in central nervous system lymphoma patients. Clin Cancer Res.2009;15(19):5968–73.

29. Wu XB, Cao DL, Zhang X, Jiang BC, Zhao LX, Qian B, et al. CXCL13/CXCR5enhances sodium channel current density via p38 MAP kinase in primarysensory neurons following inflammatory pain. Sci Rep. 2016;6:34836.

30. Jiang BC, Cao DL, Zhang X, Zhang ZJ, He LN, Li CH, et al. CXCL13 drivesspinal astrocyte activation and neuropathic pain via CXCR5. J Clin Invest.2016;126(2):745–61.

31. Strong JA, Xie W, Coyle DE, Zhang JM. Microarray analysis of rat sensoryganglia after local inflammation implicates novel cytokines in pain. PLoSOne. 2012;7:e40779.

32. Narayan K, Dail D, Li L, Cadavid D, Amrute S, Fitzgerald-Bocarsly P, et al. Thenervous system as ectopic germinal center: CXCL13 and IgG in lymeneuroborreliosis. Ann Neurol. 2005;57(6):813–23.

33. Laurent C, Dorothee G, Hunot S, Martin E, Monnet Y, Duchamp M, et al.Hippocampal T cell infiltration promotes neuroinflammation and cognitivedecline in a mouse model of tauopathy. Brain. 2017;140(1):184–200.

34. Muller G, Hopken UE, Lipp M. The impact of CCR7 and CXCR5 on lymphoidorgan development and systemic immunity. Immunol Rev. 2003;195:117–35.

35. Burkle A, Niedermeier M, Schmitt-Graff A, Wierda WG, Keating MJ, Burger JA.Overexpression of the CXCR5 chemokine receptor, and its ligand, CXCL13 inB-cell chronic lymphocytic leukemia. Blood. 2007;110:3316–25.

36. Legler DF, Loetscher M, Roos RS, Clark-Lewis I, Baggiolini M, Moser B. B cell-attracting chemokine 1, a human CXC chemokine expressed in lymphoidtissues, selectively attracts B lymphocytes via BLR1/CXCR5. J Exp Med. 1998;187(4):655–60.

37. Kim CH, Rott LS, Clark-Lewis I, Campbell DJ, Wu L, Butcher EC.Subspecialization of CXCR5+ T cells: B helper activity is focused in agerminal center-localized subset of CXCR5+ T cells. J Exp Med. 2001;193(12):1373–81.

38. Chapman KZ, Ge R, Monni E, Tatarishvili J, Ahlenius H, Arvidsson A, et al.Inflammation without neuronal death triggers striatal neurogenesiscomparable to stroke. Neurobiol Dis. 2015;83:1–15.

39. Stuart MJ, Corrigan F, Baune BT. Knockout of CXCR5 increases thepopulation of immature neural cells and decreases proliferation in thehippocampal dentate gyrus. J Neuroinflammation. 2014;11:31.

40. Su R, Sun P, Zhang D, Xiao W, Feng C, Zhong L. Neuroprotective effect ofmiR-410-3p against sevoflurane anesthesia-induced cognitive dysfunction inrats through PI3K/Akt signaling pathway via targeting C-X-C motifchemokine receptor 5. Genes Genomics. 2019;41(10):1223–31.

41. Liu S, Liu X, Xiong H, Wang W, Liu Y, Yin L, et al. CXCL13/CXCR5 signalingcontributes to diabetes-induced tactile allodynia via activating pERK,pSTAT3, pAKT pathways and pro-inflammatory cytokines production in thespinal cord of male mice. Brain Behav Immun. 2019;80:711–24.

42. Zhu QL, Luo Y, Xue QS, Zhang FJ, Yu BW. Different doses of sevofluranefacilitate and impair learning and memory function through activation ofthe ERK pathway and synthesis of ARC protein in the rat hippocampus.Brain Res. 1678;2018:174–9.

43. Li GF, Li ZB, Zhuang SJ, Li GC. Inhibition of microRNA-34a protects againstpropofol anesthesia-induced neurotoxicity and cognitive dysfunction viathe MAPK/ERK signaling pathway. Neurosci Lett. 2018;675:152–9.

44. Feld M, Krawczyk MC, Sol Fustiñana M, Blake MG, Baratti CM, Romano A,et al. Decrease of ERK/MAPK overactivation in prefrontal cortex reversesearly memory deficit in a mouse model of Alzheimer's disease. J AlzheimersDis. 2014;40(1):69–82.

45. Alam A, Hana Z, Jin Z, Suen KC, Ma D. Surgery, neuroinflammation andcognitive impairment. EBioMedicine. 2018;37:547–56.

46. Cibelli M, Fidalgo AR, Terrando N, Ma D, Monaco C, Feldmann M, et al. Roleof interleukin-1beta in postoperative cognitive dysfunction. Ann Neurol.2010;68(3):360–8.

47. Saeed M, Ghadiri A, Hadizadeh F, Attaranzadeh A, Alavi MS, Etemad L.Cinnamaldehyde improves methamphetamine-induced spatial learning andmemory deficits and restores ERK signaling in the rat prefrontal cortex. IranJ Basic Med Sci. 2018;21(12):1316–21.

48. Li Z, Liu F, Ma H, White PF, Yumul R, Jiang Y, et al. Age exacerbates surgery-induced cognitive impairment and neuroinflammation in Sprague-Dawleyrats: the role of IL-4. Brain Res. 1665;2017:65–73.

49. Garcia-Curran MM, Hall AM, Patterson KP, Shao M, Eltom N, Chen K, et al.Dexamethasone attenuates hyperexcitability provoked by experimentalfebrile status epilepticus. eNeuro. 2019;6(6) ENEURO.0430-19.2019:1-17.

50. Dube CM, Zhou JL, Hamamura M, Zhao Q, Ring A, Abrahams J, et al.Cognitive dysfunction after experimental febrile seizures. Exp Neurol. 2009;215(1):167–77.

51. Cao X, Li W, Liu Y, Huang H, Ye CH. The anti-inflammatory effects of CXCR5in the mice retina following ischemia-reperfusion injury. Biomed Res Int.2019;2019:3487607.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Shen et al. Journal of Neuroinflammation (2020) 17:335 Page 16 of 16